Non-stationary vibration road spectrum reconstruction of wheeled vehicles based on FCM clustering wavelet packet coefficient matrix

In this paper, a frequency domain non-stationary vibration spectrum simulation reconstruction method is proposed, which based on Fuzzy C-Means (FCM) clustering wavelet packet coefficient matrix reconstruction. The factors such as the non-stationarity of the road surface excitation and vibration response of the vehicle due to the time variation of the vehicle speed and the nonlinearity of the coupled vibration of the vehicle body during the running of the wheeled vehicle are considered. In this method, the one-dimensional measured vibration spectrum reconstruction problem is turned into the two-dimensional wavelet packet coefficient matrix (image) reconstruction problem by wavelet packet transform. Furthermore, based on the FCM clustering principle of wavelet packet coefficient normalized column vector, some non-stationary features in the frequency domain was extracted to realize the frequency domain non-stationary reconstruction of the measured road spectrum, including the important vibration types contained in the measured vibration spectrum and the total probability of occurrence and the adjacent conditional probability, and the dispersion around the cluster center in each type. Through the comparison with the measured road spectrum, the method was verified that it can accurately simulate the actual road driving conditions of the wheeled vehicle. The reconstructed road spectrum this method provided is consistent with the actual working conditions for road simulation experiments.